True Fantasy Points: 2016 Running Back Projections

A stats-based guide for separating skill from luck in running back projections.

As was the case with my True Fantasy Points (TFP) article on this season's quarterbacks, I'm not going to spend precious space explaining the metholodogy here. If you're interested in that sort of thing, you can click on the link above or this one. Suffice it to say that the overall concept involves using regression to the mean to figure out a player's "true" stat rates (i.e., account for how consistent particular rates are over time), and then applying those "true" rates to the amount of volume David Dodds projects for said player.

Because we're discussing running backs today, the stat rates we're relying on are related to rushing and receiving and include the following:

  • Yards per Carry (YPC)
  • Touchdowns per Carry (RuTD%)
  • Receptions per Route Run (RPRR)
  • Yards per Route Run (YPRR)
  • Touchdowns per Route Run (TDPRR)
If you're new to my TFP work, first, thanks for reading! Second, if you're unfamiliar with routes run (RR), it's a stat tracked by Pro Football Focus that both Chase Stuart and I have found to be a more reliable indicator than targets with respect to quantifying "receiving opportunities." And that's true -- pun intended -- for receptions, yards, and touchdowns, so I use RR-based rates for all three stats.1

Regarding the accuracy of running back TFP, I'll reiterate what I wrote in the 2016 quarterbacks article: 1) After only one season of implementing this system, it's too early to tell; and 2) There were hits and misses, both in the overrated and underrated direction, but the general trend was that hit rate increased as one progressed towards the extremes of the TFP difference table -- as it should.
 
All of that said, there's a 900-pound complicating factor that affects these two positions differently, so the comparison isn't entirely apples to apples: Volume itself is less stable for running backs. For a starting quarterback, reaching expected volume overwhelmingly comes down to avoiding injury and/or ineffectiveness. Except for the most marginal starters, it's rare for a quarterback that's healthy and effective through last week to see a drastic reduction in his usage this week. For running backs, in contrast, preserving one's current volume means avoiding a whole host of factors in addition to injury and/or ineffectiveness: game plans, game scripts, down/distance/field position, "hot hand" coaching whimsy, etc. Look no further than the New England and New Orleans offenses over the past decade for quintessential real-world examples of why, although one can calculate "true" rates with equal confidence for both positions, the projection of "true" opportunity is far more uncertain for running backs than quarterbacks.
 
OK, enough curtain jerking, let's proceed to the midcard.
 

TRUE STATS THROUGH 2015

Below are the actual stat rates and the "true" stat rates for the 66 running backs who are a) playing for the same team as last season, and b) projected for more than 100 "opportunities" this season -- defined as rushing attempts plus routes run (sorted by True YPC ranking):

 Actual RatesTrue Rates
PlayerTmYPCRkRuTD%RkRPRRRkYPRRRkTDPRRRkYPCRkRuTD%RkRPRRRkYPRRRkTDPRRRk
Jamaal Charles KC 5.47 5 3.18% 22 15.2% 29 1.31 26 1.07% 11 4.82 1 3.09% 17 15.2% 30 1.30 24 0.89% 2
Adrian Peterson MIN 4.90 8 4.07% 14 12.8% 47 1.04 43 0.27% 40 4.67 2 3.82% 1 13.0% 58 1.07 59 0.38% 65
Justin Forsett BAL 4.94 7 2.59% 27 14.4% 35 0.80 53 0.00% 45 4.47 3 2.79% 34 14.7% 38 0.98 66 0.38% 64
Thomas Rawls SEA 5.65 3 2.72% 25 10.5% 55 0.89 51 1.17% 9 4.47 4 2.87% 24 13.7% 52 1.20 44 0.61% 19
Jonathan Stewart CAR 4.53 20 2.81% 24 12.5% 50 1.01 44 0.51% 30 4.44 5 2.84% 30 12.8% 59 1.07 60 0.54% 41
C.J. Anderson DEN 4.75 13 3.85% 16 12.3% 51 1.06 42 0.42% 32 4.44 6 3.22% 11 13.1% 57 1.15 50 0.52% 47
Todd Gurley LA 4.83 11 4.37% 10 14.4% 36 1.29 27 0.00% 45 4.43 7 3.28% 9 14.9% 36 1.29 27 0.50% 52
Spencer Ware KC 5.60 4 8.33% 1 12.9% 45 0.11 65 0.00% 45 4.42 8 3.43% 7 14.8% 37 1.13 53 0.54% 38
LeGarrette Blount NE 4.65 15 4.23% 11 6.7% 62 0.55 61 0.56% 28 4.42 9 3.38% 8 10.9% 65 1.01 63 0.57% 27
Jerick McKinnon MIN 4.90 9 1.21% 49 20.6% 10 1.32 25 0.43% 31 4.42 10 2.57% 54 18.4% 9 1.30 25 0.54% 39
Ryan Mathews PHI 5.04 6 5.61% 6 18.5% 17 1.35 24 0.92% 15 4.41 11 3.28% 10 16.6% 21 1.30 26 0.60% 20
Chris Thompson WAS 6.00 1 0.00% 56 21.2% 7 1.38 23 1.55% 5 4.41 12 2.75% 41 18.5% 7 1.32 23 0.72% 8
Mike Gillislee BUF 5.68 2 6.38% 5 14.5% 34 0.70 55 0.00% 45 4.41 13 3.13% 15 15.2% 29 1.21 40 0.55% 36
Christine Michael SEA 4.90 10 0.00% 56 6.9% 61 0.60 57 0.00% 45 4.40 14 2.56% 57 13.6% 54 1.20 43 0.55% 37
Marcel Reece OAK 4.61 17 1.64% 41 16.0% 27 1.57 18 0.94% 14 4.40 15 2.63% 51 15.9% 25 1.52 9 0.77% 6
Darren McFadden DAL 4.56 18 1.26% 48 14.8% 31 1.22 31 0.00% 45 4.40 16 2.47% 61 15.0% 33 1.25 33 0.45% 58
David Johnson ARI 4.65 14 6.40% 4 16.7% 24 2.13 4 1.86% 2 4.40 17 3.46% 6 16.1% 23 1.63 4 0.79% 5
Darren Sproles PHI 4.61 16 6.43% 3 19.5% 12 1.59 16 0.21% 44 4.40 18 3.52% 3 18.4% 8 1.47 13 0.45% 57
DeAngelo Williams PIT 4.54 19 5.50% 7 10.5% 56 0.96 46 0.00% 45 4.40 19 3.50% 5 12.0% 61 1.11 55 0.42% 62
Dion Lewis NE 4.78 12 4.08% 13 18.6% 16 2.01 8 1.03% 12 4.39 20 2.98% 22 17.1% 14 1.56 8 0.64% 14
Tevin Coleman ATL 4.51 21 1.15% 50 2.3% 65 0.16 64 0.00% 45 4.39 21 2.70% 46 10.9% 64 1.04 62 0.53% 44
Andre Ellington ARI 4.40 24 2.47% 30 17.3% 23 1.58 17 0.52% 29 4.38 22 2.75% 39 16.9% 18 1.48 12 0.55% 33
LeSean McCoy BUF 4.41 23 1.48% 44 13.0% 42 1.19 35 0.81% 19 4.38 23 2.57% 53 14.0% 48 1.24 38 0.61% 18
Doug Martin TB 4.38 25 2.30% 33 13.9% 39 1.14 37 0.26% 41 4.38 24 2.56% 55 14.2% 45 1.18 46 0.44% 60
Kenjon Barner PHI 4.43 22 0.00% 56 23.0% 3 0.56 59 0.00% 45 4.38 25 2.79% 35 16.8% 19 1.20 42 0.55% 34
Shane Vereen NYG 4.26 33 0.00% 56 19.2% 13 1.61 15 1.30% 8 4.38 26 2.66% 49 17.8% 13 1.45 14 0.73% 7
Benny Cunningham LA 4.31 26 2.67% 26 17.5% 19 1.51 21 0.23% 43 4.38 27 2.86% 27 16.9% 17 1.41 19 0.47% 56
Cameron Artis-Payne CAR 4.07 39 2.22% 36 17.3% 21 2.01 7 0.00% 45 4.37 28 2.86% 26 15.7% 28 1.35 22 0.55% 32
Zach Zenner DET 3.53 56 0.00% 56 12.2% 52 0.67 56 0.00% 45 4.37 29 2.83% 31 15.1% 32 1.25 32 0.56% 30
Jacquizz Rodgers CHI 2.93 64 0.00% 56 5.7% 63 0.57 58 0.00% 45 4.37 30 2.84% 29 14.5% 41 1.25 36 0.56% 31
Terrance West BAL 3.91 48 0.00% 56 9.9% 57 0.52 62 0.00% 45 4.37 31 2.72% 44 14.3% 42 1.20 45 0.55% 35
Jeremy Hill CIN 4.31 27 4.49% 9 13.0% 43 0.91 50 0.31% 37 4.37 32 3.54% 2 13.8% 51 1.09 57 0.51% 49
Jay Ajayi MIA 3.82 50 2.04% 38 9.2% 58 1.18 36 0.00% 45 4.37 33 2.85% 28 13.5% 55 1.26 30 0.53% 43
Ameer Abdullah DET 4.17 35 1.40% 45 14.7% 33 1.07 41 0.59% 24 4.37 34 2.64% 50 15.0% 34 1.21 41 0.57% 25
KaDeem Carey CHI 4.01 42 2.53% 28 12.7% 48 1.21 33 1.59% 4 4.37 35 2.86% 25 14.6% 39 1.27 29 0.63% 16
Latavius Murray OAK 4.28 30 2.30% 34 14.2% 37 0.92 49 0.00% 45 4.37 36 2.70% 47 14.6% 40 1.08 58 0.41% 63
Dexter McCluster TEN 3.98 44 1.05% 53 16.0% 26 1.29 28 0.56% 26 4.36 37 2.67% 48 15.8% 26 1.28 28 0.57% 26
Fitzgerald Toussaint PIT 2.33 66 0.00% 56 0.0% 66 0.00 66 0.00% 45 4.36 38 2.83% 32 14.3% 43 1.23 39 0.56% 29
Charles Sims TB 4.13 38 0.58% 55 19.1% 14 2.05 5 1.09% 10 4.36 39 2.43% 63 17.9% 11 1.70 3 0.70% 9
Giovani Bernard CIN 4.28 31 2.44% 31 18.9% 15 1.71 12 0.64% 22 4.36 40 2.71% 45 18.3% 10 1.59 6 0.60% 21
Chris Johnson ARI 4.15 36 1.53% 43 5.1% 64 0.49 63 0.00% 45 4.36 41 2.59% 52 11.1% 62 1.06 61 0.51% 48
James White NE 3.03 63 6.45% 2 21.0% 9 2.02 6 1.87% 1 4.36 42 3.06% 19 18.5% 6 1.59 7 0.79% 4
Le'Veon Bell PIT 4.29 28 2.94% 23 15.8% 28 1.44 22 0.31% 36 4.36 43 2.92% 23 15.7% 27 1.40 21 0.45% 59
Tim Hightower NO 3.91 49 4.17% 12 21.8% 6 2.35 2 0.00% 45 4.36 44 3.06% 18 17.0% 16 1.45 15 0.54% 40
C.J. Spiller NO 3.11 62 0.00% 56 24.4% 2 1.72 11 1.44% 7 4.36 45 2.76% 38 19.5% 4 1.42 18 0.67% 11
Eddie Lacy GB 4.29 29 3.21% 21 13.6% 40 1.23 30 0.84% 17 4.36 46 3.06% 20 14.0% 47 1.24 37 0.68% 10
T.J. Yeldon JAX 4.07 40 1.10% 52 12.6% 49 0.98 45 0.35% 34 4.35 47 2.52% 59 13.6% 53 1.14 52 0.52% 45
James Starks GB 4.25 34 1.62% 42 14.9% 30 1.24 29 0.56% 27 4.35 48 2.32% 64 15.0% 35 1.25 35 0.57% 28
Carlos Hyde SF 4.06 41 3.54% 19 10.7% 54 0.56 60 0.00% 45 4.35 49 3.05% 21 12.7% 60 0.99 65 0.47% 55
Mark Ingram II NO 4.27 32 3.48% 20 16.2% 25 1.09 40 0.00% 45 4.35 50 3.21% 12 16.0% 24 1.15 49 0.36% 66
Charcandrick West KC 3.96 45 2.50% 29 8.1% 59 0.87 52 0.41% 33 4.35 51 2.83% 33 11.1% 63 1.10 56 0.54% 42
Shaun Draughn SF 3.46 59 1.32% 47 17.3% 22 1.21 32 0.00% 45 4.35 52 2.74% 42 16.3% 22 1.26 31 0.50% 51
Rashad Jennings NYG 4.15 37 1.93% 39 17.5% 20 1.55 20 0.30% 38 4.35 53 2.56% 56 16.8% 20 1.42 17 0.50% 50
Duke Johnson Jr CLE 3.64 54 0.00% 56 17.9% 18 1.56 19 0.59% 25 4.34 54 2.51% 60 17.0% 15 1.43 16 0.57% 23
Javorius Allen BAL 3.75 51 0.73% 54 22.8% 4 1.79 10 1.01% 13 4.34 55 2.53% 58 19.4% 5 1.48 11 0.64% 15
Theo Riddick DET 2.90 65 1.39% 46 25.6% 1 2.26 3 1.52% 6 4.33 56 2.76% 37 22.9% 1 1.87 1 0.86% 3
Jeremy Langford CHI 3.63 55 4.05% 15 13.0% 44 1.64 14 0.59% 23 4.33 57 3.11% 16 14.2% 46 1.41 20 0.57% 24
Devonta Freeman ATL 3.95 46 3.64% 17 21.1% 8 1.65 13 0.82% 18 4.32 58 3.15% 13 19.6% 3 1.51 10 0.65% 13
Isaiah Crowell CLE 3.94 47 3.60% 18 7.9% 60 0.76 54 0.28% 39 4.32 59 3.14% 14 10.3% 66 1.00 64 0.50% 53
Bilal Powell NYJ 4.00 43 1.74% 40 14.0% 38 1.12 38 0.25% 42 4.32 60 2.47% 62 14.2% 44 1.17 48 0.43% 61
Matt Jones WAS 3.40 61 2.08% 37 14.7% 32 2.36 1 0.78% 21 4.31 61 2.76% 36 15.1% 31 1.60 5 0.59% 22
Danny Woodhead SD 3.67 53 2.28% 35 21.9% 5 1.89 9 1.63% 3 4.31 62 2.75% 40 20.7% 2 1.71 2 1.00% 1
Melvin Gordon SD 3.48 57 0.00% 56 20.5% 11 1.19 34 0.00% 45 4.30 63 2.28% 66 17.9% 12 1.25 34 0.49% 54
Frank Gore IND 3.72 52 2.31% 32 11.9% 53 0.94 48 0.35% 35 4.30 64 2.74% 43 13.2% 56 1.12 54 0.52% 46
Mike Tolbert CAR 3.46 60 5.12% 8 13.4% 41 1.12 39 0.80% 20 4.28 65 3.51% 4 13.9% 50 1.17 47 0.65% 12
Alfred Blue HOU 3.48 58 1.14% 51 12.8% 46 0.95 47 0.85% 16 4.25 66 2.29% 65 13.9% 49 1.14 51 0.62% 17

As s a quick example of how to read this table, let's focus on the row related to Jamaal Charles. On the left side of the table are his fantasy-relevant stat rates across all of his carries and routes run with the Chiefs since 2008. The right side of the table consists of what Charles' "true" rates are after applying regression to the mean. So, while Charles' 1,320 carries have resulted in 5.47 YPC, combining a) YPC's awful predictability from one game to the next with b) the average fantasy-relevant running back having a much lower YPC than 5.47 leads to c) a 4.82 True YPC for Charles heading into 2016.

The comparison between actual and "true" rates can be interpreted in the same way for the rest of Charles' row and the rest of the running backs in the table writ large.

A few moments perusing the table reveals a couple of overarching themes. First, as I hinted at by using Charles' stats in the above example, the massive unpredictability of YPC over time means that, while actual YPC ranges from Fitzgerald Toussaint's 2.33 to Chris Thompson's 6.00, True YPC has a much narrower range: from 4.25 to 4.82. Recent work by ESPN's Mike Clay ($) suggests the following conclusion may need slightly more nuance, but the lesson here is that, in general, it's better to assume league-average YPC for a given running back than it is to assume he'll sustain his actual YPC in future games with his current team.

The second somewhat-hidden theme in the "true" side of the table is that a) rushing rate rankings positively correlate with each other (i.e., when one's high, the other's high; when one's low, the other's low); and b) so do receiving rate rankings; but c) rushing rate rankings negatively correlate with receiving rate rankings (i.e., when rushing rates are high, receiving ranks are low; and vice versa). All of which is to say -- in English this time -- that the stats back up our intuition: When drafting running backs this year, understand that the vast majority have specific "true" skill at either rushing or receiving; only a small minority have both.

To highlight a few examples of this phenomenon before moving onto my TFP projections, check out Theo Riddick's "true" stat rankings. Now check out Danny Woodhead's. Now check out Adrian Peterson's, Todd Gurley's, and LeGarrette Blount's. Sure, these are obvious examples of running backs with specific situational skill sets, but that's the point! It's so easy to rattle off specialists ad infinitum in today's NFL; much harder to name more than a handful of running backs with "truly" elite all-around skill. For fantasy purposes, the bottom line is this: In the entire table, there isn't a single running back that ranks in the Top 12 across all five "true" stat categories. Even if we expand that to the Top 24, only two running backs qualify: David Johnson and Dion Lewis.2

And now, from the midcard, to the main event.

TRUE PROJECTIONS FOR 2016

The table below displays David Dodds' standard-scoring stat projections as of August 23rd, my TFP projections based on multiplying "true" rates by his projections of opportunity volume (i.e., rush attempts and routes run), and the difference between the two (i.e., the "DIFF" column, which is how the table is sorted):

 DODDS STATSTRUE STATSPOINTS
PLAYERTMRUYDSRUTDREYDSRETDRUYDSRUTDREYDSRETDDODDSTFPDIFFRK
Todd Gurley LA 1246 12 302 2 1240 9 310 1 238.8 217.3 +21.5 1
Le'Veon Bell PIT 810 7 336 2 785 5 357 1 168.6 152.5 +16.1 2
Mark Ingram II NO 978 9 375 2 1001 7 360 1 201.3 187.0 +14.3 3
Devonta Freeman ATL 875 8 459 3 929 7 430 2 199.4 187.5 +11.9 4
David Johnson ARI 1034 10 460 3 1033 8 505 2 227.4 217.2 +10.2 5
Latavius Murray OAK 924 8 300 1 961 6 296 1 176.4 168.0 +8.4 6
Thomas Rawls SEA 968 8 208 1 961 6 236 1 171.6 163.9 +7.7 7
Jerick McKinnon MIN 450 3 176 1 442 3 155 1 86.6 79.1 +7.5 8
Frank Gore IND 689 7 251 1 732 5 262 1 142.0 134.7 +7.3 9
Darren McFadden DAL 225 2 164 1 220 1 167 1 56.9 49.7 +7.2 10
DeAngelo Williams PIT 572 6 218 1 571 5 240 1 121.0 113.9 +7.1 11
Jamaal Charles KC 1019 8 374 3 1012 6 413 3 205.3 198.2 +7.1 12
LeGarrette Blount NE 600 8 55 0 664 5 75 0 113.5 106.8 +6.7 13
LeSean McCoy BUF 956 8 273 2 986 6 347 2 182.9 178.3 +4.6 14
Giovani Bernard CIN 602 4 432 2 610 4 417 2 139.4 134.9 +4.5 15
Andre Ellington ARI 231 2 145 1 241 2 149 1 55.6 51.4 +4.2 16
Theo Riddick DET 263 2 485 3 303 2 465 2 104.8 101.2 +3.6 17
Rashad Jennings NYG 615 5 200 1 652 4 212 1 117.5 113.9 +3.6 18
Darren Sproles PHI 207 1 320 2 198 2 319 1 70.7 67.1 +3.6 19
Terrance West BAL 330 3 56 0 328 2 67 0 56.6 53.5 +3.1 20
C.J. Spiller NO 144 1 113 1 153 1 109 1 37.7 35.1 +2.6 21
Justin Forsett BAL 484 3 237 1 492 3 214 1 96.1 94.0 +2.1 22
Duke Johnson Jr CLE 573 4 473 2 630 4 437 2 140.6 139.0 +1.6 23
Shane Vereen NYG 240 1 468 3 263 2 446 2 94.8 94.0 +0.8 24
Spencer Ware KC 570 4 86 0 531 4 91 0 89.6 89.6 0.0 25
James Starks GB 489 3 272 1 501 3 268 1 100.1 100.1 0.0 26
Tim Hightower NO 336 3 154 1 371 3 170 1 73.0 73.5 -0.5 27
Christine Michael SEA 378 3 75 0 396 2 88 0 63.3 64.6 -1.3 28
C.J. Anderson DEN 1012 8 284 1 1020 7 306 1 183.6 185.4 -1.8 29
Chris Johnson ARI 336 3 61 0 371 2 77 0 57.7 60.1 -2.4 30
Bilal Powell NYJ 460 2 291 2 496 3 287 1 99.1 101.8 -2.7 31
Charcandrick West KC 239 2 108 0 239 2 129 1 46.7 50.0 -3.3 32
Dexter McCluster TEN 105 0 218 1 109 1 211 1 38.3 41.6 -3.3 33
Danny Woodhead SD 390 2 479 3 409 3 473 3 116.9 120.5 -3.6 34
Zach Zenner DET 234 2 76 0 262 2 83 0 43.0 47.0 -4.0 35
Jonathan Stewart CAR 935 6 142 1 955 6 166 1 149.7 153.8 -4.1 36
Chris Thompson WAS 320 1 332 2 353 2 293 2 83.2 87.4 -4.2 37
Jacquizz Rodgers CHI 173 1 135 1 197 1 155 1 42.8 47.0 -4.2 38
Kenjon Barner PHI 336 2 49 0 351 2 50 0 50.5 54.8 -4.3 39
Eddie Lacy GB 1131 8 200 1 1133 8 231 1 187.1 191.8 -4.7 40
Doug Martin TB 1046 7 218 1 1117 7 233 1 174.4 179.5 -5.1 41
Mike Gillislee BUF 172 1 70 0 176 1 80 0 30.2 35.3 -5.1 42
Mike Tolbert CAR 219 2 158 1 257 2 169 1 55.7 60.9 -5.2 43
Jay Ajayi MIA 620 5 179 1 677 4 216 1 115.9 121.2 -5.3 44
T.J. Yeldon JAX 672 4 253 1 697 4 267 1 122.5 127.9 -5.4 45
Cameron Artis-Payne CAR 294 2 98 0 306 2 112 0 51.2 56.6 -5.4 46
Carlos Hyde SF 989 7 224 1 1001 7 256 1 169.3 175.1 -5.8 47
Adrian Peterson MIN 1100 10 212 1 1167 10 240 1 197.2 203.0 -5.8 48
Ryan Mathews PHI 870 6 225 1 883 7 235 1 151.5 157.7 -6.2 49
Fitzgerald Toussaint PIT 300 2 60 0 327 2 69 0 48.0 54.2 -6.2 50
Benny Cunningham LA 308 1 246 1 328 2 234 1 67.4 73.7 -6.3 51
Jeremy Hill CIN 934 8 88 0 983 8 87 0 150.2 157.2 -7.0 52
Melvin Gordon SD 889 5 234 1 969 5 224 1 148.3 155.3 -7.0 53
Alfred Blue HOU 280 1 36 0 297 2 41 0 37.6 44.8 -7.2 54
Marcel Reece OAK 78 0 158 1 88 1 190 1 29.6 36.8 -7.2 55
Ameer Abdullah DET 759 4 320 2 808 5 322 2 143.9 151.4 -7.5 56
KaDeem Carey CHI 266 2 63 0 306 2 78 0 44.9 52.7 -7.8 57
Dion Lewis NE 294 1 225 1 307 2 228 1 63.9 71.7 -7.8 58
Shaun Draughn SF 293 1 203 1 326 2 209 1 61.6 70.9 -9.3 59
Tevin Coleman ATL 498 3 178 1 526 3 228 1 91.6 101.8 -10.2 60
Isaiah Crowell CLE 770 6 151 0 820 6 175 1 128.1 140.5 -12.4 61
James White NE 400 2 232 1 436 3 248 1 81.2 94.2 -13.0 62
Javorius Allen BAL 316 1 154 0 347 2 153 1 53.0 66.1 -13.1 63
Matt Jones WAS 722 5 158 1 798 5 223 1 124.0 137.6 -13.6 64
Jeremy Langford CHI 729 5 238 1 779 6 289 1 132.7 147.5 -14.8 65
Charles Sims TB 424 2 415 2 480 3 474 2 107.9 123.1 -15.2 66

This table is pretty self-explanatory, so let's cut to the chase. Here are six (or seven) running backs worth discussing based on their Dodds-TFP difference (or lack thereof):

OVERRATED: LE'VEON BELL (AND DEANGELO WILLIAMS)

As someone who's more on the stats side of stats-scouting spectrum, the love for Bell kind of puzzles me. If you look back at the "true" stats table, you see that, unlike other running backs that we just assume are at the top of the heap skill-wise (e.g., Adrian Peterson, Todd Gurley, David Johnson, etc.), Bell has a middling-at-best "true" stat profile: His highest ranking in any of the five categories was 21st. And let's not forget that Bell has the 7th-highest number of carries and the 5th-highest number of routes run among this group. In other words, sample size isn't much of an issue here.

The major discrepancies between Dodds and TFP come from three sources:

  1. He projects Bell to increase his actual 4.29 YPC to 4.50 YPC, whereas Bell's True YPC is 4.36.
  2. He projects Bell to increase his actual 2.94 RuTD% to a 3.89 RuTD%, whereas Bell's True RuTD% is 2.92%.
  3. He projects Bell to increase his actual 0.31% TDPRR to a 0.79% TDPRR, whereas Bell's True TDPRR is 0.45%.
Bell may still only be 24 years old (and that's a plus!), but we've seen enough of him from Age 21 to Age 23 -- both rushing and receiving -- to have a good gauge of his "true" efficiency, and that gauge leans more towards equilibrium than explosion.
 
As for Williams, the major divergence between Dodds and TFP comes via RuTD%, where Williams' actual rate with the Steelers is an unsustainable 5.50%. Dodds does project some regression (4.62%), but given his own average of 3.13% for these 66 running backs, and the fact that Williams only has 200 rushing attempts in his curent offense, more regression is in order -- all the way down to Williams' 3.50 True RuTD%.
 

OVERRATED: MARK INGRAM

This one's mainly about the two "true" touchdown rates. The stabilization point3 for RuTD% is 667 carries; Ingram's run the ball 748 times for the Saints. Rather than regressing Ingram's actual 3.48 RuTD% through 2015 towards the aforementioned 3.13% average, Dodds actually increases it to 3.91% for 2016. TFP, in contrast, decreases it to 3.21%.

In terms of TDPRR, Ingram has yet to score a receiving touchdown in his five-year career, which encompasses 635 routes run for the Saints. Is this a fluke? In his favor, the stabilization point for TDPRR is 1,063 routes run, so this zero-touchdown performance is only about 40 percent of the way towards reflecting "true" going-for-some-kind-of-anti-record skill. That said, given Ingram's projection of 312 routes run, 2 receiving touchdowns seems a tad optimistic as it would require improving from the stone cold worst (0.00%) all the way to average (0.64%). TFP is more conservative, projecting a TDPRR almost smack dab in between those two values: 0.36%.

OVERRATED: DEVONTA FREEMAN

Freeman's TFP-related weaknesses are like Ingram's, except with an additional layer of overratedness (if that's even a word). Not only do Freeman's "true" stats suggest that he'll score fewer rushing and receiving touchdowns than Dodds projects, they also suggest he'll gain fewer receiving yards. That's because, given his 488 routes run with the Falcons last season, Freeman's actual 1.65 YPRR should be regressed down to 1.51, whereas Dodds projects only a slight dip down to 1.61.

underrated: CHARLES SIMS

Dodds underestimates Sims in three of the four stat categories, but the most glaring differences come in the yardage department. Through 173 carries with Tampa Bay, he has an actual 4.13 YPC. This is below the 4.38 average for all 103 fantasy-relevant running backs in my 2016 database, the 4.44 average for the 66 running backs in the table, and even the 4.21 average Dodds projects for the running backs in the table. Coupling this with the unpreditability of the stat means that Sims' Actual YPC currently represents 8 percent skill and 92 percent luck. And therefore, we should be heavily regressing his below-average YPC upwards toward whichever of the above means we prefer. That's what TFP does, from 4.13 to 4.36, but Dodds does the opposite, from 4.13 to 3.85.

As for receiving yardage, this time both of us anticipate a drop in Sims' 5th-ranked actual YPRR (2.05), but Dodds' drop is much larger (down to 1.49) than is TFP's (down to 1.70). Only nine other running backs in the table have a bigger discrepancy between the two systems' expectations of how much YPRR will change this season.

UNDERRATED: JEREMY LANGFORD

One of those nine running backs is Langford. And indeed, he ends up having the same yardage-based profile as Sims for explaining his Dodds-TFP difference. TFP projecting 51 more receiving yards arises from a 0.25-yard difference between Langford's 1.41 True YPRR and what Dodds projects. Meanwhile, the 50-yard difference in his rushing projection is due to Langford benefitting from TFP's more favorable YPC adjustment. Namely, TFP regresses his abysmal 3.63 actual YPC all the way up to 4.36 True YPC, whereas Dodds' adjustment is less sanguine (4.05).

UNDERRATED: JAMES WHITE

Earlier, I mentioned how Dion Lewis was one of only two running backs to rank in the Top 24 of all five "true" stat categories. Well, if we restrict ourselves to only the three receiving categories (i.e., RPRR, YPRR, and TDPRR), it turns out that only four rank in the Top 12 of all three: The earlier-mentioned Theo Riddick and Danny Woodhead; the just-mentioned Charles Sims; and the guy who happens to be replacing Lewis in the Patriots' starting lineup, James White.

What's more, among the 12 most underrated running backs per the "DIFF" column, the only starters underrated in all four elements of their fantasy projection are White and the just-mentioned Jeremy Langford.  Here's White's TFP breakdown:

  • White's Actual YPC is an unsustainably low 3.03. Dodds projects an increase to 4.00. TFP assumes a True YPC of 4.36.
  • White's Actual RuTD% is an unsustainably high 6.45%. Dodds projects a decrese to 2.00%. TFP assumes a True RuTD% of 3.06%.
  • White's Actual YPRR is an above-average 2.02. Dodds projects a decrease to 1.48. TFP assumes a True YPRR of 1.59.
  • White's Actual TDPRR is an unsustainably high (and top-ranked, in fact) 1.87%. Dodds projects a decrease to 0.64%. TFP assumes a True TDPRR of 0.79%.
Trepidation among fantasy owners about White typically revolves around taking a wait-and-see approach due to his miniscule sample size of carries and routes run. First off, I'm glad you're thinking about sample size. Getting to the heart of the matter, though, White's is exactly the type of projection that benefits from using my TFP system. That's because adjusting for sample sizes is precisely what it does. TFP "knows" White only has 31 carries and 214 routes run with the Patriots. But it also "knows" that rushing stats are more random than receiving stats, and that White's performance across 214 routes run has been as great as his performance across 31 carries has been awful. Put this together and we have the underlying reason for White being underrated: While what he seems to be great at requires a small sample-size adjustment downward, what he's awful at requires a massive sample-size adjustment upward.

HONORABLE MENTIONS

Underrated: Matt Jones and Isaiah Crowell.

Overrated: Todd Gurley and Frank Gore.

Just Right: Duke Johnson Jr, C.J. Anderson, Justin Forsett, Spencer Ware, and James Starks.

1 The RR data I used for 2007-2014 are actual totals published by Pro Football Focus before they removed the actual totals from their site. Beginning last season, they've only provided "per-opportunity" stats from which one must extrapolate RR. It's eminently doable; you just end up with an estimated RR for 2015 (and beyond), and thus there's a tiny bit of measurement error introduced into the TFP numbers.

2 Hint: What I'm saying here is that, barring some statistically outlying event -- or series of events -- Johnson is the real deal...in all scoring formats...in all lineup formats...redraft or dynasty...Seriously.

3 The stabilization point is the threshold for a stat representing 50% skill and 50% randomness. For instance, the stabilziation point for RuTD% is 667 carries. When a running back has run the ball exactly that many times, his RuTD% gives us 50% of the information we need to determine his "true" rushing touchdown skill. Before 667 carries, RuTD% is mostly a reflection of luck. After 667 carries, RuTD% is mostly a reflection of skill.


More articles from Danny Tuccitto

See all

More articles on: Projections

See all

More articles on: RB

See all

More articles on: Stats

See all